OpenAI makes GPT-5.6 Luna the free default and removes text-chat limits

OpenAI rolled out a broad ChatGPT update that resets the free experience around GPT-5.6 Luna, now the default for free and Go users, paired with unlimited text chats and a new 'Think' button that surfaces more deliberate reasoning on demand. OpenAI says Luna cuts responses containing at least one factual error by 62%, and in a high-stakes factuality eval spanning finance, medicine and law the refreshed GPT-5.6 Sol produced 68% fewer error-containing responses than GPT-5.5 Instant.

Paid Plus and Pro users get the updated Sol powering all chats, including Instant, for a single consistent experience, plus a 'thinking effort' slider — Simon Willison noted you can long-press the mobile send button to adjust it. OpenAI also added dedicated U18 evaluations and teen safeguards, and positioned the agentic ChatGPT Work suite at the center of an enterprise productivity push. The changes arrive as ChatGPT crosses 1 billion weekly users.

Community reaction split along the paywall. Developers cheered an ~80% price cut on Luna for high-volume tasks, but some argued the free tier now delivers a 'quality gap' by locking free users out of OpenAI's stronger reasoning models. Others debated whether a more accurate but lighter Luna is a fair trade for losing access to deeper reasoning.

Competitively, the move undercuts free tiers from Google's Gemini and Anthropic's Claude while OpenAI simultaneously paused its more capable Astra model on safety grounds — a striking split between aggressive consumer expansion and frontier caution. Watch adoption metrics, whether the factuality gains hold up outside OpenAI's own evals, and how rivals respond on free-tier limits.